GET Trolling Detector: How We Increased Productivity Using a Speech Analyzer / Mail.ru Group Blog / Sudo Null IT News FREE

In the modern world, corporate trolling is even as common as spam. In the rating of "time eaters", trolls are in the first place, and the point is not only when that there are too many of them divorced, but also that, unlike spam (against which very telling tools have already appeared), corporate trolls mimic for indifferent office workers, and topics for troll are settled on what, in fact, is the bring. Therefore, it is sometimes very vexed to signalize the employee's simple desire to "troll" from the substantive oil-bearing dispute.

The Mail.Ru Group internal resources efficiency department has found a right smart to distinguish troll from productive communication: we created the AntiTroll hardware-software complex, tested it and made for sure of its effectiveness. Details under the cut.

First, one cannot break to bring up that communication with the troll reduces the effectualness of not only the troll itself (which, the likes of a swamp, is becoming wider with fresh water pouring into it), but likewise of the employees who participate in the discussion. Therefore, a decrease in effectiveness is often of an epidemiological (viral) nature.

In ordination to determine the level of troll, we plunged into dianoetic analysis, studied the oral communicatio used by our team members (we wrote about the terms used by our developers here .

During this analysis, both the form of the language and its function were considered. As a reference The material was taken in colloquial words and written texts in which we identified the linguistic features of the sensing of various texts and types of oral speech.

The analysis of written texts enclosed the study of the maturation of the topic and the linguistics connection (coupling) between sentences. The guys from Mail.Ru Search helped us a lot therein counselling.

The analysis of colloquial speech was based both along the above aspects and on the practices of stone's throw-by-step fundamental interaction, opening and closure sequences of social interactions surgery the structure of story.

The working group developing the AntiTroll mathematical setup had to seriously study psycholinguistics, sociolinguistics, sociology, hermeneutics, social psychology and even anthropology (as it turned out, such parameters as gender, height and acoustic properties of the articulative (speech) setup should too be taken into account when processing information )

In parallel with the development of the mathematical apparatus, we accumulated data for analysis.
Our usability laboratory was associated to this. After a poor period of adaption, the experimental chemical group of the game development department employees started talking about their public project, and with the help of a decoy trolling, the discussion moved from the stage of effective communicating to the phase of low-efficiency troll connected a traditional screenland topic (for example, Android vs iOS surgery Windows vs Linux) .

Gigabytes of information were collected past the registrars of pulse, respiration, activity of the cerebral cortex, microphones and video recorders. This initial information was fed to a script longhand on the groundwork of the above mathematical apparatus.

The result of the script should represent a individualist number - the storey of trolling.

After 28 builds of the script, we were close to abandoning this musical theme: the script produced some strange values, rather resembling the result of the random number generator. And then we decided to add other weighting transmitter to the mathematical model: "take down of passionate involution". Information technology is precisely called the "personal appraisal of existing operating theatre possible situations." And suddenly, on the 29th assembly, "it" worked!

The script began to determine the troll level quite accurately, and when somebody affected from the stage of productive discourse to the degree of verbal dilatoriness, the numerical value of the trolling level steadily grew.

Terminated the past 7 months, we proven this system in 5 departments of Mail.Ru Group, gradually adjusting the script, and at the moment we cause achieved the accuracy of crucial trolling in 96%.

It is interesting that during testing, an analogue of the "Ballmer peak" , known to most habrayuzers, was unconcealed. Busy a certain value, troll near does not bear on productivity, but when the value of X is reached, developers have a sharp increase in productivity (the number of commits increases, the number of bugs decreases). However, when threshold X is exceeded, productivity drops sharply.

Whether this uncovering has practical application is not yet comprehensible: information technology is obvious that it is quite difficult to maintain the level of trolling in the play team at the required dismantle to achieve this peak.

The next step was the creative activity of the AntiTroll Alarm! Computer hardware and software program complex:

The principle is quite simple: a microphone located in the meeting way transmits an audio signal to a constantly running server with AntiTroll book, and if a overflowing level of troll is detected, then to Arduino, connected to the server via USB , a signal comes on that the "AntiTroll Alarm!" relay is on. As before long as the trolling level in the discussions rolls over, a siren and a flashing insignificant start up:

While we tested our new system in different units for three months, we already see positive results: the amoun of commits increased by 23%, the number of bugs establish by testers attenuated by 14%.

In the draw near future, we plan to introduce the system in all divisions, and then completely make believe mobile applications that john be put-upon not only by us, but also by colleagues in the market.

Just imagine - in front the planning session, put the phone in the center of the table, become on our application, and Don't worry approximately missing the consequence when your discussion goes into a non-constructive centering ...

The seed cypher leave later be posted on github here .

DOWNLOAD HERE

GET Trolling Detector: How We Increased Productivity Using a Speech Analyzer / Mail.ru Group Blog / Sudo Null IT News FREE

Posted by: touchstonethiphrisity.blogspot.com

Share this post